MemeCheckShieldShort backtest note vs /rugs
Published for grant M1/M2 honesty. Numbers come only from stored study rechecks in the evidence tracker — never invented.
منشور لشفافية المنحة. الأرقام من إعادات فحص الدراسة المخزّنة فقط — لا اختراع.
Risk-level backtest (study_v2)
As of 2026-10-20T03:11:16Z, completed 24h rechecks: n = 57.
| Level at scan | n | RUGGED % | Notes |
|---|---|---|---|
| HIGH | 2 | 100% (2/2) | Both High flags ($NODF, $TikTok US) rugged afterwards — tiny sample, wide CI |
| MEDIUM | — | 69.7% | From overall 24h table on /rugs |
| LOW | — | 86.4% | Most fresh memes in sample died anyway |
Risk-level backtest vs /rugs (same entry rules). High→RUGGED catch rate and Low/Medium baseline rug rates are published from stored rechecks only. Dev Reputation Score creators were not stored in the tracker for this cohort, so a formal DevRep-label backtest is not claimed yet; tracker will start recording creator wallets going forward.
What this means for Dev Reputation
- v1 DevRep uses the same dead/rugged rules as Developer history (launches, survival, cadence, graduations, LP sample ≤5).
- This cohort did not store creator wallets in
db.json, so we do not claim a formal Builder/Serial/High-Risk label backtest yet. - Going forward the tracker will record
creatorwhen available so DevRep labels can be scored against the same outcomes. - Live methodology: /how#devrep · lookup: /dev
Machine-readable: accuracy-daily.json → backtest field.
اختبار رجعي لمستوى الخطر
حتى 2026-10-20T03:11:16Z، إعادات فحص 24 ساعة مكتملة: n = 57.
- High → سحب: 100% (n=2) — عينة صغيرة جدًا
- Medium سحب: 69.7% · Low سحب: 86.4%
اختبار رجعي لمستوى الخطر مقابل /rugs (نفس قواعد الدخول). ننشر معدل اصطياد High→RUGGED ومعدلات السحب لـ Low/Medium من إعادات الفحص المخزّنة فقط. محافظ منشئ Dev Reputation لم تُحفظ في هذه الدفعة، لذلك لا ندّعي اختبارًا رجعيًا رسميًا لتسمية DevRep بعد؛ سيبدأ المتعقّب بتسجيل محافظ المنشئ لاحقًا.